Download Measuring the Chirp and the Linewidth Enhancement Factor of

Measuring the Chirp and the Linewidth
Enhancement Factor of Optoelectronic
Devices with a Mach–Zehnder Interferometer
Volume 3, Number 3, June 2011
Jean-Guy Provost
Frederic Grillot, Member, IEEE
DOI: 10.1109/JPHOT.2011.2148194
1943-0655/$26.00 ©2011 IEEE
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
Measuring the Chirp and the Linewidth
Enhancement Factor of Optoelectronic
Devices with a Mach–Zehnder
Interferometer
Jean-Guy Provost 1 and Frederic Grillot,2;3 Member, IEEE
1
III-V Lab, a joint lab of Alcatel-Lucent Bell Labs France, Thales Research and Technology,
and CEA Leti, 91460 Marcoussis, France
2
Université Européenne de Bretagne, CNRS FOTON, Institut National des Sciences
Appliquées, 35043 Rennes Cedex, France
3
Institut TELECOM/Telecom Paristech, CNRS LTCI, 75634 Paris Cedex, France
DOI: 10.1109/JPHOT.2011.2148194
1943-0655/$26.00 !2011 IEEE
Manuscript received April 20, 2011; accepted April 21, 2011. Date of publication April 29, 2011; date
of current version May 20, 2011. Corresponding author: J.-G. Provost (e-mail: jean-guy.provost@
3-5lab.fr).
Abstract: In this paper, a technique based on the use of a Mach–Zehnder (MZ) interferometer is proposed to evaluate chirp properties, as well as the linewidth enhancement factor
(!H -factor) of optoelectronic devices. When the device is modulated, this experimental setup
allows the extraction of the component’s response of amplitude modulation (AM) and frequency modulation (FM) that can be used to obtain the value of the !H -factor. As compared
with other techniques, the proposed method gives also the sign of the !H -factor without
requiring any fitting parameters and, thus, is a reliable tool, which can be used for the
characterization of high-speed properties of semiconductor diode lasers and electroabsorption modulators. A comparison with the widely accepted fiber transfer function method is
also performed with very good agreement.
Index Terms: Chirp, electroabsorption modulators, linewidth enhancement factor, optical
modulation, semiconductor lasers.
1. Introduction
The linewidth enhancement factor (!H -factor) is used to distinguish the behavior of semiconductor
lasers (SLs) with respect to other types of lasers [1] and influences several fundamental aspects,
such as the linewidth [1], [2], the chirp under modulation [3], the laser’s behavior under optical
feedback [4], [5], as well as the occurrence of the filamentation in broad-area lasers [6]. The
!H -factor is usually defined as the coupling between the phase and the amplitude of the electric field
such as [1]
!H ¼ "
4" dn=dN
4"! dn=dN
¼"
# dg=dN
# dGnet =dN
(1)
where # is the lasing wavelength, N is the carrier density, g is the material gain, ! is the optical
confinement factor, and Gnet ¼ !g " !i is the net modal gain with !i the internal loss coefficient. The
!H -factor depends on the ratio of the evolution of the refractive index n with the carrier density N to
that of the differential gain dg=dN. As reported in [7], several different techniques have been
proposed to measure the !H -factor, with no rigorous comparison between the results achieved, as
Vol. 3, No. 3, June 2011
Page 476
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
pointed out in [8]. Also, it should be stressed that the number of the proposed measuring methods
has kept increasing while novel types of SLs, such as those based on quantum dot (QD), have
arisen, for which the determination of the !H -factor may be particularly critical [9].
In most cases, the !H -factor is evaluated by using the so-called Hakki–Paoli subthreshold
method, which relies on direct measurement of the refractive index change and the differential gain
as the carrier density is varied by slightly changing the current of an SL [10], [11]. This method,
which is applicable only below threshold, gives the material !H -factor and does not correspond to
an actual lasing condition. As a consequence of that, it makes more sense to determine the
!H -factor above the laser’s threshold. Thus, relevant aspects such as the high-power behavior of
the !H -factor due to nonlinear effects and the consequences on the adiabatic chirp can be taken
into account. Consequently, the !H -factor appears as an optical power-dependent parameter that is
strongly influenced by nonlinear gain and/or carrier heating effects [12]. Such a power-dependence
is particularly strengthened in QD lasers in which the lasing wavelength can switch from the ground
state (GS) to the excited state (ES) as the injected current increases. This accumulation of carrier in
the ES arises, even though lasing in the GS that is still occurring enhances the effective !H -factor of
the GS transition introducing a nonlinear dependence with the injected current [9]. Among the
above-threshold techniques used for extracting the !H -factor, the linewidth method relies on the
measurement of SL’s linewidth, as well as on fitting the results to known SL’s parameters [13]–[15].
Two other possibilities are based on injection-locking or on optical feedback techniques. On one
hand, light from a master SL is injected into the slave SL, causing locking of the slave optical
frequency to that of the master and an asymmetry in frequency due to the nonzero !H -factor [16],
[17]. On the other hand, the optical feedback method is based on the self-mixing interferometry
configuration from which the !H -factor is determined from the measurement of specific parameters
of the resulting interferometric waveform [18]. Also, we can note that the !H -factor of a distributed
feedback (DFB) SL has been measured by using a phase-controlled high-resolution optical lowcoherence reflectometer [19]. Finally, the determination of the !H -factor can be conducted through
high-frequency techniques. These methods are much more relevant when the performances of the
device operating under direct modulation have to be evaluated for high-speed transmissions. On
one hand, the SL current modulation generates both amplitude (AM) and optical frequency (FM)
modulation [20]. The ratio of the FM over AM components gives a direct measurement of the
!H -factor [20]–[23]. The AM term can be measured by direct detection via a high-speed photodiode,
while the FM term is related to sidebands intensity that can be measured using a high-resolution
scanning Fabry–Perot filter. Although the FM/AM method requires modulation well above the SL’s
relaxation frequency, this technique gives the device !H -factor under direct modulation. On the
other hand, the fiber transfer function method originally proposed for the electroabsorption modulators (EAMs) [24] exploits the interaction between the chirp of a high-frequency modulated SL and
the chromatic dispersion of an optical fiber, which produces a series of minima in the amplitude
transfer function versus modulation frequency. Such a technique has then been generalized to
diode lasers by introducing the adiabatic term, as shown in [25] and [26], and by fitting the
measured transfer function, the !H -factor can be retrieved. This method has been shown to be
reliable, as long as precise measurement of fiber dispersion is made and as long as the power
along the fiber span is kept enough low to avoid nonlinear effects. As compared with the FM/AM
technique, the main disadvantage of such a method is that several fitting parameters have to be
determined to access the !H -factor.
Another important issue concerns the determination of the sign of the !H -factor, which is of first
importance for many applications requiring ultralow laser linewidth, such as on-chip pulse compression, chirp compensation, or for the EAMs. Among all the techniques explained above, only the
fiber transfer function method can give the phase and then the sign of the !H -factor. In this paper,
an optical discriminator based on a tunable Mach–Zehnder (MZ) interferometer is used to extract
AM and FM responses both in amplitude and in phase in the frequency domain as well as the
!H -factor. In Section 2, the basic equations needed to extract the !H -factor, as well as the experimental setup, are presented. Section 3 shows experimental results on the !H -factor of DFB lasers,
EAMs, and integrated laser modulators (ILMs), as well as a comparison with measurements
Vol. 3, No. 3, June 2011
Page 477
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
Fig. 1. Experimental setup used for the determination of AM, FM responses, and of the !H -factor.
(Inset) Transfer function of the interferometer.
obtained with the fiber transfer method. Finally, we summarize our results in Section 4. Although both
Michelson and MZ interferometers have already been used in the past to measure the SL’s FM
responses [27]–[30], to the best of our knowledge, extractions of the chirp-to-power ratio (CPR) and
of the !H -factor have not been reported yet. As pointed out in [31], the FM and AM responses, as well
as the !H -factor, have been measured in the time domain through the MZ interferometer. Although
such a technique remains very efficient when transient and adiabatic chirp contributions have to be
separated [32], it does not lead to the same level of performances. Thus, because of the equipment
limitations (PRBS generator with fixed transition time, sampling oscilloscope), the sensitivity, the
dynamic, and the accuracy of the method is not as good as the one proposed in this paper. Let us
stress that impact of the thermal effects is much more complicated to evaluate with large-signal
measurements because low-frequency operation is required. Consequently, this interferometric
technique is to be of first importance to measure the high-speed properties of the next generation of
lasers and modulators.
2. Experimental Setup and Theory
2.1. Experimental Description
As depicted in Fig. 1, our experimental setup is similar to the one originally developed by
Sorin et al. [33]. The goal is to determine the characteristics of the FM induced by the current
modulation in the laser’s cavity or the phase modulation (PM) for an external modulator. The signal at
the output is analyzed through a tunable MZ interferometer made with two fibered couplers. This
interferometer has a free-spectral range (FSR), which is the inverse of the differential delay jT2 " T1 j
between the two arms, T1 and T2 being the propagation time (time delay) in the two arms, respectively. A polarization controller is used to make sure that the two signals located at the input of the
second coupler have parallel states. The inset in Fig. 1 shows the power at the output of the
interferometer as a function of the propagation time difference or of the optical frequency. To
accurately control the optical path difference, a cylindrical piezoelectric transducer is used. The
transducer located onto one of the MZ’s arms is fiber interdependent and directly controlled by an
external locking circuit. The system allows adjustment of the interferometer on all points of the
characteristics. For instance, points A and B being in opposition, they correspond to two signals
interfering in quadrature with each other, as shown in the inset of Fig. 1. Around these two locations,
the interferometer’s characteristics remaining linear, the photocurrent coming out from the
Vol. 3, No. 3, June 2011
Page 478
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
photodetector is proportional to the phase (or frequency) variations of the optical signal to be
analyzed.
2.2. Theoretical Description of a Laser Under Direct Modulation (AM and FM)
The electric field from a laser under direct modulation can be expressed as follows:
pffiffiffiffiffi
eðt Þ ¼ P0 ð1 þ m cosð2"fm t ÞÞ1=2 &exp j ½2"f0 t þ $ sinð2"fm t þ ’Þ(
(2)
with P0 the average power; m the modulation rate in power ðm ¼ =P0 Þ; fm the frequency of the
electrical signal provided by the network analyzer; f0 the central optical frequency; $ the modulation
rate in frequency, such as $ ) "F =fm ("F being the amplitude of the FM across the optical carrier
f0 ); and ’ the phase difference between the modulation frequency and the amplitude frequency. At
the output of the interferometer, the signal can be written as
1
sðt Þ ¼ ½eðt " T1 Þ þ eðt " T2 Þ(:
2
(3)
By injecting (2) into (3)
sðt Þ ¼
1 pffiffiffiffiffi
P0 ½1 þ m cosð2"fm ðt " T1 ÞÞ(1=2 exp j ½2"f0 ðt " T1 Þ þ $ sinð2"fm ðt " T1 Þ þ ’(
2
1 pffiffiffiffiffi
þ
P0 ½1 þ m cosð2"fm ðt " T2 ÞÞ(1=2 exp j ½2"f0 ðt " T2 Þ þ $ sinð2"fm ðt " T2 Þ þ ’(: (4)
2
As mentioned in Section 2.1., measurements are done at two different points A and B in opposition
and corresponding to two signals interfering in quadrature with each other, meaning that
2"f0 ðT2 " T1 Þ ¼ ð"=2Þ þ k ", with k being an integer. Assuming this condition, the photocurrent
going toward the network analyzer being proportional to sðt Þs* ðt Þ can be expressed as follows:
$
"
#
"
"
##%
P0
"fm
T1 þ T2
1 þ m cos
sðt Þs* ðt Þ ¼
cos 2"fm t "
2
FSR
2
"
"
#
"
"
##
P0
"fm
T1 þ T2
þ"
1 þ 2m cos
cos 2"fm t "
2
FSR
2
#1=2
þ m 2 cosð2"fm ðt " T1 ÞÞ cosð2"fm ðt " T2 ÞÞ
"
"
#
"
"
#
##
"fm
T1 þ T2
& sin 2$ sin
cos 2"fm t "
þ’ :
FSR
2
(5)
In (5), " is a parameter whose value (+1) depends on the position A or B, as depicted in the inset of
Fig. 1, while FSR ¼ 1=jT1 " T2 j is the FSR of the MZ interferometer. Also, it should be stressed that
the network analyzer being sensitive only to the signal’s components beating at the FM, all the
continuous and higher order terms such as ð2fm ; 3fm ; . . .Þ can be neglected [33]. Considering these
assumptions, (5) can be rewritten as follows:
$
"
#
"
"
##%
P0
"fm
T1 þ T2
*
m cos
sðt Þs ðt Þ ¼
cos 2"fm t "
2
FSR
2
"
"
##
"
"
#
#
"fm
T1 þ T2
þ "P0 J1 2$ sin
cos 2"fm t "
þ’
(6)
FSR
2
with J1 the Bessel function of the first kind, which can be approximated in most case such as
J1 ð2$sinð"fm =FSRÞÞ , $sinð"fm =FSRÞ. As a consequence of that, (6) can be simplified and
Vol. 3, No. 3, June 2011
Page 479
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
rewritten such as
$
"
#
"
"
##%
P0
"fm
T1 þT2
mcos
sðt Þs ðt Þ ¼
cos 2"fm t "
2
FSR
2
*
"
#
"
"
#
#
"fm
T1 þ T2
þ "P0 $sin
cos 2"fm t "
þ ’ : (7)
FSR
2
As shown in Fig. 1, the network analyzer giving two results associated with the A and B operating
points, respectively, the normalized measured signals can be expressed as follows:
"
#
"
#
P0
"fm
"fm
m cos
M+ ¼
expð"j2"fm % Þ + P0 $ sin
exp j ð"2"fm % þ ’Þ
2
FSR
FSR
(8)
where Mþ is the result in A and M" in B, respectively, while % ¼ ðT1 þ T2 Þ=2 is the transit time within
the interferometer. On one hand, the first term in (8) only depends on the AM (i.e., $-independent),
while the term cosð"fm =FSRÞ corresponds to the AM transfer function of the interferometer [33]. On
the other hand, the second term in (8) is purely related to the modulation frequency (m independent)
and can be expressed as a function of the interferometer FM transfer function, such as
"P0 ð"F =FSRÞsincðfm = FSRÞ with sincðx Þ ¼ sinð"x Þ="x .
From (8), following expressions can be deduced:
(
(
(M þ " M " (
2$
1
(
(
& "f ' (
¼
m
m
Mþ þ M" (
tan FSR
"
#
Mþ " M"
’ ¼ arg
:
Mþ þ M"
(9a)
(9b)
Using the definitions of parameters m and $, (9a) allows extraction of the CPR [34], such as
(
(
(Mþ " M" (
"F
fm
1
(
(:
& "f ' (
¼
m
"P 2P0 tan FSR
Mþ þ M" (
(10)
The value of the !H -factor is then determined through the so-called relationship [21]
2$
¼ !H
m
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
" #2
fc
1þ
:
fm
(11)
In (11), fc is defined as the corner frequency [35]
fc ¼
1
@g
vg
P
2" @P
(12)
with vg the group velocity, P the output power, and @g=@P a nonzero parameter because of the
phenomenon of nonlinear gain related to nonzero intraband relaxation times, as well as carrier
heating. Parameter @g=@P can be expanded as a function of the gain compression factor "P
following the relationship [36]:
@g
"P g
¼
:
@P 1 þ "P P
(13)
For typical numbers, the corner frequency can be in the hundreds of Megahertz to the few
Gigahertz range, depending on the output power level. On one hand, for modulation frequencies
such as fm - fc , which is the case in the experiment since the maximum modulation frequency fm is
Vol. 3, No. 3, June 2011
Page 480
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
Fig. 2. Amplitude (solid line) and Phase (dotted line) of the CPR as a function of the modulation
frequency for the QW DFB laser under study.
about in the 20-GHz range, the factor 2$=m directly equals the laser’s !H -factor. On the other hand,
for lower modulation frequencies, the ratio 2$=m becomes inversely proportional to the modulation
frequency. Let us note that the measurement of 2$=m with frequency and at different output power
levels could serve for the determination of the corner frequency and, consequently, the gain
compression factor.
2.3. Theoretical Description of an External Modulator (AM and PM)
In case of an external modulator the phase variation is given by [37]
d &ðt Þ !H 1 dPðt Þ
¼
dt
2 Pðt Þ dt
(14)
with &ðt Þ the instantaneous phase of the optical signal and Pðt Þ the related power. Under a small
signal analysis condition, the optical power PðtÞ can be expressed such as
Pðt Þ ¼ P0 ð1 þ mcosð2"fm t ÞÞ:
(15)
Then, it can be shown that the signal at the output of the modulator can be written following the
relationship:
eðt Þ ¼
)
*
pffiffiffiffiffi
!H
mcosð2"fm tÞ :
P0 ð1 þ mcosð2"fm t ÞÞ1=2 expj 2"f0 t þ
2
(16)
Based on a similar calculation as the one conducted in Section 2.2, the !H -factor can be expressed as
"
#
1
1
Mþ " M"
& "f '
!H ¼
:
m
j tan FSR
Mþ þ M"
(17)
3. Experimental Results and Discussion
3.1. Case of a DFB Laser
The device under study is a DFB laser having a high reflection (HR) coating on the rear facet and
an antireflection (AR) coating on the front facet to allow for high efficiency. The device is 350 'm
long with an active layer made with quantum wells nanostructures. The threshold current is about
7.5 mA at room temperature (25 . C). In Fig. 2, the ratio "F ="P in amplitude and in phase is
depicted for a DFB laser operating under direct modulation in the range from 10 kHz to 15 GHz. The
Vol. 3, No. 3, June 2011
Page 481
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
Fig. 3. 2$=m as a function of the modulation frequency for the QW DFB laser under study.
amplitude term of "F ="P is obtained through (10), considering a 5-mW optical output power (this
value is obtained for a DC bias current equal to 2.6 times the threshold current) for the laser under
study, while the FSR equals 47.6 GHz (see the Appendix). With regard to the phase term of
"F ="P, it is obtained from (9b). At low frequencies ðfm G 10 MHzÞ, thermal effects are predominant [3], [36]. When the modulation frequency decreases, the continuous wave regime gets
closer, and the phase difference between the FM and AM responses tends to 180. [3], [36]. Such
behavior is similar to the static operation case in which an increase in the laser’s emitting
wavelength (a decrease in the laser’s optical frequency, respectively) is observed both with the
injected current and with the output power. When 10 MHz G fm G 2 GHz, thermal effects are no
longer significant, and the CPR is relatively constant. This regime corresponds to the adiabatic
regime dominated by the gain compression effects and in which the AM and FM modulations are inphase [3], [36]. Then, for larger frequencies ðfm 9 2 GHzÞ, relaxation oscillations between the
carrier and photon numbers take place. The CPR gets proportional to fm and the FM and AM
responses are in quadrature with each other, leading to a 90. phase difference. In Fig. 3, the
measured 2$=m ratio is plotted via (9a) starting from 50 MHz (beyond the thermal effects). As
predicted by (11), the function 2$=m tends asymptotically to the !H -factor, which is estimated to be
about 2.4 for the laser under study.
3.2. Case of an EAM
In Fig. 4, the EAM’s !H -factor, as well as the corresponding phase, is evaluated through (17) for a
monochromatic incident optical signal. On one hand, in Fig. 4(a), the !H -factor is measured for a
"2.0 V reverse voltage. This figure shows that the phase term remains constant and equal to zero
so that the !H -factor is positive (þ1.32 in that case). On the other hand, Fig. 4(b) shows that when
the reverse voltage decreases down to "3.2 V, the phase term drops to "180. so that the
measured !H -factor gets negative ("1.20 in that case). Those results demonstrate that phase
variations can be used to obtain the sign of the !H -factor.
Let us note that for EAM-based devices, the !H -factor remaining frequency-independent,
measurements could be conducted at one frequency only, which is much quicker by comparison
with the fiber transfer function method [24]. Indeed, to reach a good resolution on the minima of
transmission, the fiber transfer method typically requires a wider span up to 15 or 20 GHz, as well
as a lot of data points (/400). Considering the same component as well as the same wavelength,
Fig. 5 shows a comparison between the two methods as a function of the bias voltage. As it can be
shown, a very good agreement is demonstrated; in particular, the bifurcation point from which the
!H -factor switches from positive to negative values occurs at "2.6 V for both methods. However,
when the bias voltage equals "4 V, which corresponds to !H / "10, the discrepancy between the
two methods starts increasing. This discrepancy is due to the lower experimental accuracy which
typically arises when the !H -factors gets larger ðj!H j 0 10Þ, as pointed out in [24] for the fiber
transfer function method and in Section 3.4 (see below) for the method under study.
Vol. 3, No. 3, June 2011
Page 482
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
Fig. 4. Measured !H -factor and phase as a function of the modulation frequency for the EAM.
(a) V ¼ "2:0 V. (b) V ¼ "3:2 V.
Fig. 5. Comparison of the !H -factor measurements by the fiber method (blue line) and MZ method (red
line) as a function of the reverse voltage for the EAM ðfm ¼ 10 GHzÞ.
3.3. Case of an ILM
Theoretically speaking, the chirp of an ILM should be similar to the one obtained for an isolated
modulator, meaning that it should not be frequency-dependent, as already pointed out in
Section 3.2. However, in an actual situation, a perturbation whose origin comes from an electrical
or optical feedback is usually observed in the amplitude response [38], [39]. This unwanted
feedback leads to either a positive or a negative dip arising close to the relaxation frequency of the
laser section, as well as impacting the chirp behavior. Fig. 6(a) shows an example of the ILM’s
!H -factor measured for a bias current of 100 mA (for the laser section) and for a "1 V bias voltage
(for the EAM section). As shown, a dip occurring close to 7.7 GHz and with a full bandwidth of 3 GHz
is observed. Based on relative intensity noise (RIN) measurements, the laser’s relaxation frequency
is confirmed to occur at 7.7 GHz for the same bias current condition. In Fig. 6(a), the phase being
disturbed in the dip area, the !H -factor is estimated to be about þ0.17 outside this region. Such a
Vol. 3, No. 3, June 2011
Page 483
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
Fig. 6. (a) Measured !H -factor amplitude (solid line) and phase (squared line) as a function of the
modulation frequency for the ILM. (b) Corresponding AM response.
perturbation appears more pronounced on the chirp characteristic, as compared with the AM
response on which this phenomenon could be less perceptible, as shown in Fig. 6(b).
3.4. Evaluation of the !H -Factor’s Experimental Accuracy
This section aims to evaluate the experimental accuracy related to the measurement of the
!H -factor (or to the 2$=m ratio) in the case of fm ¼ 10 GHz. This last value is considered because it
corresponds to a realistic value used for the determination of the chirp parameter (see Section 3.1).
Typically, the two main sources of uncertainties occurring in the determination of the !H -factor are
those related to the accuracy of the FSR of the MZ interferometer, as well as to the linearity of the
electrooptics network analyzer. All other contributions can be neglected and have minor effects, as
compared with the overall accuracy. For instance, the electrical frequency modulation fm is known
with an accuracy of 10"5 , while we can demonstrate that the locations around the points A and B
have no effect, at the first order, on the signal measured by the network analyzer.
• Accuracy on the FSR
In the Appendix, it is shown that the FSR is equal to 47.6 + 0.2 GHz.
Consequently, for fm ¼ 10 GHz, accuracy of the term tanð"fm =FSRÞ occurring in (9a) and (17)
does not exceed 0.7%.
• Nonlinearity of the electrooptics network analyzer
First, let us note that in the case of the CPR and !H -factor measurements, the network
analyzer’s calibration is not required since the correction factor vanishes through the ratio
jðMþ " M" Þ=ðMþ þ M" Þj, which occurs in the set of (9a), (10), and (17). Consequently, only the
nonlinear behavior of the network analyzer has to be taken into account for the estimation of
the experimental accuracy. Thus, by using a referenced optical attenuator, the deviation of the
analyzer’s linearity at 10 GHz and, in the optical power operating range at the input of the
photo-detector, is found to be at most equal to 0.15 dB, which can lead to an error of 1.7% on
the ratio jMþ =M" j. In Table 1, the experimental accuracy of the !H -factor (or of the 2$=m ratio)
has been evaluated, taking into account both the nonlinearity of the analyzer, as well as the
additional contribution of the FSR. Calculations are done for different values of the !H -factor
Vol. 3, No. 3, June 2011
Page 484
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
TABLE 1
Experimental accuracy on the !H -factor taking into account both the nonlinearity of the analyzer as well
the additional contribution of the FSR. Calculations are done for different values of the !H -factor and at
fm ¼ 10 GHz
and at fm ¼ 10 GHz. Let us note that regarding the CPR [see (10)], the output power precision
(+2% in the best case [40]) has to be included in the accuracy calculations.
We also have to consider the assumption made in (7)
"
"
##
"
#
"fm
"fm
J1 2$sin
, $sin
:
FSR
FSR
For instance, for m ¼ 10% and !H ¼ 2, the calculated deviation is 0.2% for fm ¼ 10 GHz but
increases to 1.2% for !H ¼ 5. However, let us stress that such a deviation remains negligible as
long as m 1 5%.
4. Conclusion
In this paper, an optical discriminator based on a tunable MZ interferometer has been used to
extract FM/AM ratio, as well as the !H -factor for both laser diodes and EAM-based devices. As
such a method allows the determination of both modulus and phase over a wide frequency span,
more relevant information on the chirp can be extracted, as compared with the traditional
techniques like the fiber transfer method, which only holds under the assumption that the chirp is
not frequency-dependent. In case of DFB lasers, the proposed method also allows evaluation of the
adiabatic chirp and the thermal effects. With regard to the EAM, the experimental results have been
found to be in a very good agreement with those obtained from the fiber transfer. As discussed, the
proposed technique is also much quicker, as compared with the fiber transfer one and can also be
used to evaluate the influence of the optical feedback on the EAM’s laser section. Finally, let us
stress that the proposed experimental setup can also cover a wide range of operating wavelengths
since only the couplers and the optical fibers are wavelength-sensitive and can easily be converted
to a large-signal analysis configuration [31], leading to complementary results from those presented
in this paper. As compared with other techniques, this method based on a tunable MZ
interferometer requires no fitting parameters and, thus, is a reliable tool, which can be used for
the characterization of high-speed properties of semiconductor diode lasers and EAMs.
Appendix
Determination of t he Optimum Value of t he FSR of t he
MZ Interferometer
Relationships obtained in Section 2 allow determination of the optimum value of the FSR. Indeed,
the term tanð"fm =FSRÞ occurring in (9a), (10), and (17) can be a source of uncertainty especially
Vol. 3, No. 3, June 2011
Page 485
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
Fig. 7. Measurement of the transfer function of the MZ interferometer.
when the modulation frequency gets closer to half of the FSR (see Section 3.4). In our case,
measurements are conducted up to 20 GHz (corresponding to the network analyzer bandwidth
limit) so that the FSR of the interferometer has to be a bit greater than 40 GHz. As demonstrated in
Section 2.2 for the laser case, the signals measured by the network analyzer can be expressed as
follows (8):
M+ ¼
"
#
"
#
P0
"fm
"F
"fm
m cos
sin
expð"j2"fm %Þ + P0
exp jð"2"fm % þ ’Þ
fm
2
FSR
FSR
(A1)
in which the parameter $ has been replaced by its definition.
To simplify (A1), let us consider the case FSR - fm . In that case, the following approximations
can be made cosð"fm =FSRÞ , 1 and sinð"fm =FSRÞ , "fm =FSR such that (A1) becomes
M+ ¼
P0
"F
exp jð"2"fm % þ ’Þ:
m expð"j2"fm %Þ + P0 "
FSR
2
(A2)
When the FSR is too large, the second part of the right-hand side of (A2) becomes much
weaker compared with the first part. Consequently, this situation could enhance the sensitivity to
noise and to experimental evolution between measurements in Mþ and M" (small decoupling
effect, . . .).
As a conclusion, to overcome such a problem, a 50-GHz value was targeted at the time that the
interferometer was built. The interferometer transfer function is measured with a broadband optical
source and an optical spectrum analyzer (OSA) to obtain the value of the FSR. In Fig. 7, the
obtained value between the two vertical solid lines is equal to "# ¼ 3:818 nm (ten times the FSR).
A result as accurate as 47.6 + 0.2 GHz is deduced for the FSR where +0.2 GHz takes into account
both the accuracy of the OSA as well as the experimental resolution related to the position of the
minima on the interferometer’s characteristic.
Acknowledgment
The authors would like to thank D. Leclerc, who proposed to work on this subject, and J.-L. Beylat,
H. Bissessur, T. Ducelier, J. Jacquet, C. Kazmierski, C. Smith, and B. Thedrez for fruitful discussions
and encouragements.
Vol. 3, No. 3, June 2011
Page 486
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
References
[1] C. H. Henry, BTheory of the linewidth of semiconductor lasers,[ IEEE J. Quantum Electron., vol. QE-18, no. 2, pp. 259–
264, Feb. 1982.
[2] H. Su, L. Zhang, A. L. Gray, R. Wang, T. C. Newell, K. J. Malloy, and L. F. Lester, BLinewidth study of InAs-InGaAs
quantum dot distributed feedback lasers,[ IEEE Photon. Technol. Lett., vol. 16, no. 10, pp. 2206–2208, Oct. 2004.
[3] K. Petermann, Laser Diode Modulation and Noise. Norwell, MA: Kluwer, 1991.
[4] D. M. Kane and K. A. Shore, Unlocking Dynamical Diversity. Hoboken, NJ: Wiley, 2005.
[5] F. Grillot, N. A. Naderi, M. Pochet, C.-Y. Lin, and L. F. Lester, BVariation of the feedback sensitivity in a 1.55-'m InAs/
InP quantum-dash Fabry–Perot semiconductor laser,[ Appl. Phys. Lett., vol. 93, no. 19, p. 191 108, Nov. 2008.
[6] J. Marciante and G. P. Agrawal, BNonlinear mechanisms of filamentation in broad-area semiconductor lasers,[ IEEE J.
Quantum Electron., vol. 32, no. 4, pp. 590–596, Apr. 1996.
[7] M. Osinski and J. Buus, BLinewidth broadening factor in semiconductor lasersVAn overview,[ IEEE J. Quantum
Electron., vol. QE-23, no. 1, pp. 9–29, Jan. 1987.
[8] G. Giuliani, S. Donati, A. Villafranca, J. Lasobras, I. Garces, M. Chacinski, R. Schatz, C. Kouloumentas, D. Klonidis,
I. Tomkos, P. Landais, R. Escorihuela, J. Rorison, J. Pozo, A. Fiore, P. Moreno, M. Rossetti, W. Elsiisser, J. Von Staden,
G. Huyet, M. Saarinen, M. Pessa, P. Leinonen, V. Vilokkinen, M. Sciamanna, J. Danckaert, K. Panajotov, T. Fordell,
A. Lindberg, J.-F. Hayau, J. Poette, P. Besnard, F. Grillot, M. Pereira, R. Nelander, A. Wacker, A. Tredicucci, and
R. Green, BRound-robin measurements of linewidth enhancement factor of semiconductor lasers in COST 288 action,[
presented at the Conf. Lasers Electro-Optics (CLEO) Europe/Int. Quantum Electronics Conf. (IQEC), Munich, Germany,
2007, Paper CB9-2-WED.
[9] F. Grillot, B. Dagens, J. G. Provost, H. Su, and L. F. Lester, BGain compression and above threshold linewidth
enhancement factor in 1.3-'m InAs-GaAs quantum-dot lasers,[ IEEE J. Quantum Electron., vol. 44, no. 10, pp. 946–
951, Oct. 2008.
[10] B. W. Hakki and T. L. Paoli, BGain spectra in GaAs double-heterostructure injection laser,[ J. Appl. Phys., vol. 46, no. 3,
pp. 1299–1306, Mar. 1975.
[11] I. D. Henning and J. V. Collins, BMeasurements of the semiconductor laser linewidth broadening factor,[ Electron. Lett.,
vol. 19, no. 22, pp. 927–929, Oct. 1983.
[12] G. P. Agrawal, BEffect of gain and index nonlinearities on single-mode dynamics in semiconductor lasers,[ IEEE J.
Quantum Electron., vol. 26, no. 11, pp. 1901–1909, Nov. 1990.
[13] Z. Toffano, A. Destrez, C. Birocheau, and L. Hassine, BNew linewidth enhancement determination method in
semiconductor lasers based on spectrum analysis above and below threshold,[ Electron. Lett., vol. 28, no. 1, pp. 9–11,
Jan. 1992.
[14] A. Villafranca, J. A. Lázaro, I. Salinas, and I. Garcés, BMeasurement of the linewidth enhancement factor in DFB lasers
using a high-resolution optical spectrum analyzer,[ IEEE Photon. Technol. Lett., vol. 17, no. 11, pp. 2268–2270,
Nov. 2005.
[15] A. Villafranca, A. Villafranca, G. Giuliani, and I. Garces, BMode-resolved measurements of the linewidth enhancement
factor of a Fabry–Perot laser,[ IEEE Photon. Technol. Lett., vol. 21, no. 17, pp. 1256–1258, Sep. 2009.
[16] R. Hui, A. Mecozzi, A. D’Ottavi, and P. Spano, BNovel measurement technique of !-factor in DFB semiconductor lasers
by injection locking,[ Electron. Lett., vol. 26, no. 14, pp. 997–998, Jul. 1990.
[17] G. Liu, X. Jin, and S. L. Chuang, BMeasurement of linewidth enhancement factor of semiconductor lasers using an
injection-locking technique,[ IEEE Photon. Technol. Lett., vol. 13, no. 5, pp. 430–432, May 2001.
[18] Y. Yu, G. Giuliani, and S. Donati, BMeasurement of the linewidth enhancement factor of semiconductor lasers based on
the optical feedback self-mixing effect,[ IEEE Photon. Technol. Lett., vol. 16, no. 4, pp. 990–992, Apr. 2004.
[19] C. Palavicini, G. Campuzano, B. Thedrez, Y. Jaouen, and P. Gallion, BAnalysis of optical-injected distributed feedback
lasers using complex optical low-coherence reflectometry,[ IEEE Photon. Technol. Lett., vol. 15, no. 12, pp. 1683–
1685, Dec. 2003.
[20] C. Harder, K. Vahala, and A. Yariv, BMeasurement of the linewidth enhancement factor! of semiconductor lasers,[
Appl. Phys. Lett., vol. 42, no. 4, pp. 328–330, Feb. 1983.
[21] R. Schimpe, J. E. Bowers, and T. L. Koch, BCharacterization of frequency response of 1.5-'m InGaAsP DFB laser
diode and InGaAs PIN photodiode by heterodyne measurement technique,[ Electron. Lett., vol. 22, no. 9, pp. 453–454,
Apr. 1986.
[22] U. Kruger and K. Kruger, BSimultaneous measurement of the linewidth enhancement factor!, and FM and AM response
of a semiconductor laser,[ J. Lightw. Technol., vol. 13, no. 4, pp. 592–597, Apr. 1995.
[23] T. Zhang, N. H. Zhu, B. H. Zhang, and X. Zhang, BMeasurement of chirp parameter and modulation index of a semiconductor laser based on optical spectrum analysis,[ IEEE Photon. Technol. Lett., vol. 19, no. 4, pp. 227–229, Feb. 2007.
[24] F. Devaux, Y. Sorel, and J. F. Kerdiles, BSimple measurement of fiber dispersion and of chirp parameter of intensity
modulated light emitter,[ J. Lightw. Technol., vol. 11, no. 12, pp. 1937–1940, Dec. 1993.
[25] A. Royset, L. Bjerkan, D. Myhre, and L. Hafskjaer, BUse of dispersive optical fibre for characterisation of chirp in
semiconductor lasers,[ Electron. Lett., vol. 30, no. 9, pp. 710–712, Apr. 1994.
[26] R. C. Srinivasan and J. C. Cartledge, BOn using fiber transfer functions to characterize laser chirp and fiber dispersion,[
IEEE Photon. Technol. Lett., vol. 7, no. 11, pp. 1327–1329, Nov. 1995.
[27] H. Olesen and G. Jacobsen, BPhase delay between intensity and frequency modulation of a semiconductor laser
(including a new measurement method),[ in Proc. ECOC, 1982, pp. 291–295, Paper BIV-4.
[28] D. Welford and S. B. Alexander, BMagnitude and phase characteristics of frequency modulation in directly modulated
GaAlAs semiconductor diode lasers,[ J. Lightw. Technol., vol. LT-3, no. 5, pp. 1092–1099, Oct. 1985.
[29] E. Goobar, L. Gillner, R. Schatz, B. Broberg, S. Nilsson, and T. Tanbun-ek, BMeasurement of a VPE-transported DFB
laser with blue-shifted frequency modulation response from DC to 2 GHz,[ Electron. Lett., vol. 24, no. 12, pp. 746–747,
Jun. 1988.
Vol. 3, No. 3, June 2011
Page 487
IEEE Photonics Journal
Chirp and Linewidth Enhancement Factor
[30] R. S. Vodhanel and S. Tsuji, B12 GHz FM bandwidth for a 1530 nm DFB laser,[ Electron. Lett., vol. 24, no. 22,
pp. 1359–1361, Oct. 1988.
[31] R. A. Saunders, J. P. King, and I. Hardcastle, BWideband chirp measurement technique for high bit rate sources,[
Electron. Lett., vol. 30, no. 16, pp. 1336–1338, Aug. 1994.
[32] R. Brenot, M. D. Manzanedo, J.-G. Provost, O. Legouezigou, F. Pommereau, F. Poingt, L. Legouezigou, E. Derouin,
O. Drisse, B. Rousseau, F. Martin, F. Lelarge, and G. H. Duan, BChirp reduction in quantum dot-like semiconductor
optical amplifiers,[ presented at the Eur. Conf. Exh. Opt. Commun., Berlin, Germany, 2007, Paper we.08.6.6.
[33] W. V. Sorin, K. W. Chang, G. A. Conrad, and P. R. Hernday, BFrequency domain analysis of an optical FM
discriminator,[ J. Lightw. Technol., vol. 10, no. 6, pp. 787–793, Jun. 1992.
[34] T. L. Koch and J. E. Bowers, BNature of wavelength chirping in directly modulated semiconductor lasers,[ Electron. Lett.,
vol. 20, no. 25, pp. 1038–1040, Dec. 1984.
[35] L. Olofsson and T. G. Brown, BFrequency dependence of the chirp factor in 1.55 'm distributed feedback semiconductor
lasers,[ IEEE Photon. Technol. Lett., vol. 4, no. 7, pp. 688–691, Jul. 1992.
[36] L. A. Coldren and S. W. Corzine, Diode Lasers and Photonic Integrated Circuits. Hoboken, NJ: Wiley, 1995.
[37] F. Koyama and K. Iga, BFrequency chirping in external modulators,[ J. Lightw. Technol., vol. 6, no. 1, pp. 87–93,
Jan. 1988.
[38] P. Brosson and H. Bissessur, BAnalytical expressions for the FM and AM responses of an integrated laser-modulator,[
IEEE J. Sel. Topics Quantum Electron., vol. 2, no. 2, pp. 336–340, Jun. 1996.
[39] N. H. Zhu, G. H. Hou, H. P. Huang, G. Z. Xu, T. Zhang, Y. Liu, H. L. Zhu, L. J. Zhao, and W. Wang, BElectrical and
optical coupling in an electroabsorption modulator integrated with a DFB laser,[ IEEE J. Quantum Electron., vol. 43,
no. 7, pp. 535–544, Jul. 2007.
[40] D. Derickson, Fiber Optic Test and Measurement. Englewood Cliffs, NJ: Prentice-Hall, 1998.
Vol. 3, No. 3, June 2011
Page 488